由於現今少有快速的橢圓偵測演算法，而能利用硬體加速的演算法更少，但是近年來圖檔大小的快速增加，未來有些應用可能會需要能夠利用硬體加速的橢圓偵測演算法，而且未來圖檔的大小可能會無法讓整張圖輸入到硬體裝置上，所以我們在此介紹一個以GPU實作的方塊橢圓辨識器。為了使演算法更有效率和容易實作在GPU上，我們利用橢圓標準式來固定了每個橢圓組合的計算時間以及輸入的大小; 為了處理大圖檔，我們將原圖分割成數個子圖，用數個拋物線來代表每段在子圖中的線段，接著把這些介於子圖間的弧線合併。在三個Dataset共1227張圖片，與目前快速橢圓偵測演算法相比，我們的演算法有較高的F-measure或是較快的速度，而我們提出演算法GPU實作的版本比CPU的版本速度快300 至 645 %。

There are very less ellipse detectors with fast speed, and those with fast speed are seldom with hardware acceleration. However, with the size of images growing rapidly in recent years, some applications may need an ellipse detector with hardware acceleration devices someday. Furthermore, we may not load the whole image into hardware devices with huge size of image in future. As the result, we introduce an efficient block-based ellipse detector with GPU implementation. To make the algorithm more efficient and easily implemented on GPU, we fixed the time of calculation and the size of input data for each combination of ellipses candidates with ellipse equation; to deal with huge images, we split image into several subimages, use several parabola arcs to represent each original arcs in split subimages and then merge them back between subimages. In comparison with other fast ellipse detectors, our ellipse detector is faster speed or higher precision, and GPU version runs 300%-645% faster than CPU version for our ellipse detector.

[1] M. L. Balter, A. I. Chen, T. J. Maguire, and M. L. Yarmush, "Adaptive Kinematic Control of a Robotic Venipuncture Device Based on Stereo Vision, Ultrasound, and Force Guidance," IEEE Trans Ind Electron, vol. 64, no. 2, pp. 1626-1635, Feb 2017.
[2] G. Wang, J. Wu, and Z. Ji, "Single view based pose estimation from circle or parallel lines," Pattern Recognition Letters, vol. 29, no. 7, pp. 977-985, 2008.
[3] Z. Yinqiang, M. Wenjuan, and L. Yuncai, "Another way of looking at monocular circle pose estimation," IEEE International Conference on Image Processing, pp. 861-864, 2008.
[4] S. Zhao et al., "A Robust Real-Time Vision System for Autonomous Cargo Transfer by an Unmanned Helicopter," IEEE Transactions on Industrial Electronics, vol. 62, no. 2, pp. 1210-1219, 2015.
[5] M. Fornaciari, A. Prati, and R. Cucchiara, "A fast and effective ellipse detector for embedded vision applications," (in English), Pattern Recognition, vol. 47, no. 11, pp. 3693-3708, Nov 2014.
[6] Q. Jia, X. Fan, Z. Luo, L. Song, and T. Qiu, "A Fast Ellipse Detector Using Projective," arXiv, p. 14, 2016.
[7] S. Chen, R. Xia, J. Zhao, Y. Chen, and M. Hu, "A hybrid method for ellipse detection in industrial images," Pattern Recognition, vol. 68, pp. 82-98, 2017.
[8] A. Y. Chia, S. Rahardja, D. Rajan, and M. K. Leung, "A split and merge based ellipse detector with self-correcting capability," IEEE Trans Image Process, vol. 20, no. 7, pp. 1991-2006, Jul 2011.
[9] B.-K. Kwon, Z. Teng, T.-J. Roh, and D.-J. Kang, "Fast ellipse detection based on three point algorithm with edge angle information," International Journal of Control, Automation and Systems, vol. 14, no. 3, pp. 804-813, 2016.
[10] Y. Ito, K. Ogawa, and K. Nakano, "Fast Ellipse Detection Algorithm Using Hough Transform on the GPU," pp. 313-319, 2011.
[11] C. Akinlar and C. Topal, "EDCircles: A real-time circle detector with a false detection control," Pattern Recognition, vol. 46, no. 3, pp. 725-740, 2013.
[12] C. Akinlar and C. Topal, "Edpf: A Real-Time Parameter-Free Edge Segment Detector with a False Detection Control," International Journal of Pattern Recognition and Artificial Intelligence, vol. 26, no. 01, p. 1255002, 2012.
[13] G. Maier, "Optimal arc spline approximation," Computer Aided Geometric Design, vol. 31, no. 5, pp. 211-226, 2014.
[14] W. Guo and M.-J. Lai, "Box Spline Wavelet Frames for Image Edge Analysis," SIAM Journal on Imaging Sciences, vol. 6, no. 3, pp. 1553-1578, 2013.
[15] R. Achanta, A. Shaji, K. Smith, A. Lucchi, P. Fua, and S. Susstrunk, "SLIC superpixels compared to state-of-the-art superpixel methods," IEEE Trans Pattern Anal Mach Intell, vol. 34, no. 11, pp. 2274-82, Nov 2012.
[16] P. L. Rosin, "Analysing error of fit functions for ellipses," Pattern Recognition Letters, vol. 17, no. 14, pp. 1461-1470, 1996.
[17] “Cubic spline interpolation“, [Online]. Available: http://nmarkou.blogspot.tw/2012/03/cubic-spline-interpolation.html. [Accessed 6 7 2017]
[18] Wolfram Mathworld, “Ellipse”, [Online]. Available: http://mathworld.wolfram.com/Ellipse.html. [Accessed 6 7 2017]
[19] Math Forum – Ask Dr. Math, “Computing Ellipse Parameters” [Online]. Available: http://mathforum.org/library/drmath/view/51735.html. [Accessed 6 7 2017]
[20] “CUDA C Programming Guide” [Online]. Available: http://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html. [Accessed 6 7 2017]
[21] “NVIDIA’s Fermi: The First Complete GPU Computing Architecture” [Online]. Available: http://www.nvidia.com/content/pdf/fermi_white_papers/p.glaskowsky_nvidia's_fermi-the_first_complete_gpu_architecture.pdf. [Accessed 6 7 2017]
[22] “CUDA Overview” [Online]. Available: https://www.cc.gatech.edu/~vetter/keeneland/tutorial-2011-04-14/02-cuda-overview.pdf [Accessed 6 7 2017]
[23] Jens Breitbart, “A framework for easy CUDA integration in C++ applications” [Online]. Available: http://camlunity.ru/swap/Library/Conflux/Competitors/framework_for_easy_cuda_integration_c___applications.pdf [Accessed 6 7 2017]